The Blog

view all

Colombia Sensory Trial: Mapping G < (E + M)

2015-06-23 Comments Off on Colombia Sensory Trial: Mapping G < (E + M)

Two weeks ago, I introduced the G (genotype) x E (environment) x M (management) framework here, and told you about a key finding from the Colombia Sensory Trial: although we designed the Trial to examine the impact of G on cup quality, we found G to be less correlated with cupping scores than E and M.

Today I present a series of sensory maps created by researchers at Kansas State University’s Sensory Analysis Center–the ones who developed the World Coffee Research lexicon and applied it (for the first time ever) to the Castillo and Caturra samples in the Trial–to show you how G < (E + M).


WCR Lexicon - Attributes


This sensory map plots 32 references in the WCR lexicon.  Each blue dot reflects the intensity of a sensory reference for the corresponding attribute.  Each attribute is described in the lexicon with precision and the sensory references are prepared according to specific recipes and anchored to a reference intensity; the references are on the table with the coffee samples to aid in the assessment process.  Here is an example for the attribute “Fruity-Dark,” which was used as part of the Colombia Sensory Trial:

Fruity-Dark: An aromatic impression of dark fruit that is sweet and slightly brown associated with dried plums and raisins.

References: 1/4 cup Sun Maid raisins and 1/4 cup of Sun Maid prunes (chopped), 3/4 cup of water   = 6.0 (f), 5.0 (a); Diluted Sunsweet prune juice = 4.5 (f); 3(a)

Preparation: Mix raisins (whole) and prunes (chopped). Add ¾ cup of water and cook in microwave on high for 2 minutes. Filter with a sieve. Use the blender to mix the paste of raisins and prunes for 30 seconds. Serve the paste into 1 ounce cups. Place 1 TBSP of the liquid juice and the paste in a medium snifter and cover with watch glass.

Mix Sunsweet brand prune juice with water at a 50% ratio. Serve approx. 1 oz. and 1 TBSP in medium snifter. This may be prepared 24 hours in advance and refrigerated in coded, lidded sample cups. Bring to room temperature for serving.

For more than half the farms in the Trial, the Castillo and Caturra samples tasted alike.  They didn’t just get the same scores, they actually tended to taste alike.  In 13 of 22 cases, the Castillo and Caturra samples from the same farms tended to taste more like one another than two Castillos from different farms did, or two Caturras from different farms.  These 13 sample pairs were further divided into three separate clusters based on similarities in sensory attributes across farms.  Those clusters are plotted against the attribute references on the following maps.


WCR Lexicon - Cluster 1


WCR Lexicon - Cluster 2


WCR Lexicon - Cluster 3


The first and second clusters, the smallest two, represented by red and purple dots, were mapped into areas concentrated with attributes we would generally consider undesirable in our coffees.  The farms in third, and largest, cluster in the upper left-hand quadrant, was highest in attributes buyers and consumers alike would likely consider desirable, including caramelized, overall sweet and fruity berry.

The results of the Trial’s cupping panels suggested that cup scores/quality were more correlated with environment and management than variety.  The WCR lexicon analysis made a similar but different suggestion: that the specific sensory attributes of the samples (related to but distinct from cup scores/quality) were also more correlated with environment and management than variety.